Feature or enhancement

Proposal

json.dumps escapes each string by first scanning it one character at a time to compute the escaped size (ascii_escape_size in Modules/_json.c); when nothing needs escaping, write_escaped_ascii then copies the string verbatim. For a long string with no characters that need escaping, which is the common case for text values, log messages, and other long content, that per-character sizing scan is pure overhead before the verbatim copy.

The proposal is to detect the no-escape case on the one-byte (ASCII/Latin-1) representation eight bytes at a time. Load eight bytes into a single machine word and test all eight at once for a character that needs escaping (c < 0x20, c > 0x7e, c == '"', or c == '\\'). When a long run has none, return the verbatim size directly. A length guard keeps short strings, such as the typical dict key, on the existing per-character loop, where the eight-byte path would not pay for its setup. Strings that need escaping, and two-byte and four-byte strings (anything with a non-Latin-1 character), keep the current path.

This is the encode-side counterpart to the decode-side scan in #150871 (PR #150872). The two touch different code paths, so they are separate changes.

How this differs from the SIMD backend in #142915

It is not the SIMD parsing architecture declined in #142915. It uses no SIMD intrinsics, no runtime CPU detection, and no build configuration, only portable 64-bit integer arithmetic with the same 0x0101… / 0x8080… masks that Objects/unicodeobject.c already applies for ASCII scanning. It changes one function and adds no infrastructure, so it does not depend on #125022 and needs no PEP.

When it helps, and when it does not

Measured json.dumps speedups against the current encoder:

The benefit is specific to documents whose payload is long, escape-free text. The short-string guard keeps key-heavy documents unaffected.

Correctness

The encoded output is byte-identical to the current encoder. A patch is validated against test_json and a 199-case differential corpus (strings placing each escape-relevant character, including ", \, control chars, 0x7f, and non-Latin-1 characters, at every offset across the eight-byte window, in both ensure_ascii=True and ensure_ascii=False modes). Every output matched.

A proof-of-concept PR follows.

Benchmark

Built base and patched interpreters from this branch's main ancestor and the patch, ran the same script under each, and compared with pyperf compare_to (A/B by swapping Lib/json/encoder.py on the same build; macOS arm64, non-PGO).

import json, pyperf
long_ascii = [("x"*200) for _ in range(200)]                 # long no-escape ASCII values
text_blob  = {"body": "lorem ipsum dolor sit amet " * 400}   # one huge no-escape string
escaped    = [('a"b\\c\n'*30) for _ in range(200)]           # escape-heavy
short_keys = {f"k{i}": i for i in range(2000)}               # short keys
mixed_real = [{"id":i,"name":f"user_{i}","email":f"u{i}@x.com","bio":"hello world "*10} for i in range(300)]
nonascii   = ["café 😀 中文 "*20 for _ in range(200)]          # UCS-2/4 (scalar path)
objs={"long_ascii":long_ascii,"text_blob":text_blob,"escaped":escaped,"short_keys":short_keys,"mixed_real":mixed_real,"nonascii":nonascii}
r=pyperf.Runner()
for n,o in objs.items():
    r.bench_func(f"dumps/{n}", lambda o=o: json.dumps(o))

Linked PRs

• gh-150875: Speed up JSON string encoding for long ASCII strings #150876